Journal of Chromatography A, 984 (2003) 89–96

www.elsevier.com / locate / chroma

E

valuation of headspace solid-phase microextraction for the analysis

of volatile carbonyl compounds in spirits and alcoholic beverages

*

´

Waldemar Wardencki , Piotr Sowinski, Janusz Curyl«o

´

Department of Analytical Chemistry

, Chemical Faculty, Technical University of Gdansk, 11 /12 Narutowicz Str.,

´

80-952 Gdansk, Poland

Received 2 August 2002; received in revised form 21 October 2002; accepted 21 October 2002

Abstract

A method was developed for the determination of C –C carbonyl compounds in alcoholic solutions using pentafluoro-

1

6

benzoxymation followed by headspace sampling solid-phase microextraction and subsequent analysis by GC with electron-
capture detection. Experimental conditions—alcohol content, exposure time, temperature and sample agitation were
optimised. In this method, a spirit or distilled alcoholic beverage is first adjusted to 20% (v / v) alcohol. Detection limits for
particular aldehydes and ketone varied from 0.05 to 0.5 mg / l and relative standard deviation was between 2.3 and 20%.
Generally, the method showed good linearity for the tested concentration range 8 mg / l–0.32 mg / l with regression
coefficients ranging between 0.9434 and 0.9983. The method was applied to the analysis of real alcoholic beverages
(vodkas).



2002 Published by Elsevier Science B.V.

Keywords

: Food analysis; Derivatisation, GC; Carbonyl compounds; Aldehydes

1

. Introduction

responsible for unpleasant organoleptic properties of
alcoholic drinks. Furthermore, they can bind in vivo

There is still a great demand for reliable and

to biological nucleophiles (proteins, DNA, cellular

sensitive methodology for determination of individ-

membranes, enzymes) resulting in toxic mutagenic

ual carbonyl compounds (aldehydes and ketones) in

and carcinogenic effects.

different foodstuffs and drinks due to their potential

Problems with carbonyl compounds determina-

adverse health effects [1,2].

tions are caused by: (a) their low concentrations and

In spirits and alcoholic beverages, low molecular

complex matrices (in alcoholic matrices up to 15

mass carbonyls (C –C ) are present as by-products

carbonyls may be present), (b) wide range of their

1

6

of yeast fermentation, intermediates in the formation

concentration (from mg / l to ng / l levels) and (c) by

of fusel oil and as a result of alcohol oxidation at

their high reactivity, especially in the case of unsatu-

various stages of beverage production. Their pres-

rated constituents.

ence is undesirable because some of them are

Carbonyl species can be determined by several

methods (titrimetric, colorimetric, enzymatic) but
HPLC and GC are the most convenient techniques.

*Corresponding author. Fax: [phone].

E-mail address

:

[e-mail]

(W. Wardencki).

The application range of these techniques may be

0021-9673 / 02 / $ – see front matter



2002 Published by Elsevier Science B.V.

P I I : S 0 0 2 1 - 9 6 7 3 ( 0 2 ) 0 1 7 4 1 - 7

90

W

. Wardencki et al. / J. Chromatogr. A 984 (2003) 89–96

expanded by transformation of these compounds into

of 37% aqueous solution. Dimethylketone and hexa-

appropriate derivatives and using selective detection.

nal were obtained from POCH (Gliwice, Poland).

Four common derivatization agents, i.e. 2,4-dinitro-

Stock standard solutions of each carbonyl compound

phenylhydrazine

(DNPH)

[3,4],

2,4,6-trichloro-

were prepared in 10 ml of methanol (J.T. Baker,

phenylohydrazine (TCPH) [5], cysteamine (2-amino-

Deventer, Netherlands) and then diluted to the

ethanethiol)

[6,7]

and

O-(2,3,4,5,6-pentafluoro-

required concentration with neutral ethanol in the

benzyl)hydroxyamine (PFBHA) [8–12] have been

concentration range of 8 and 320 mg / l. The de-

found useful for this purpose. The derivatives formed

rivatizing reagent PFBHA (Sigma–Aldrich) was

(hydrazones, thiazolidines and oximes) prior to chro-

dissolved in doubly distilled water at a concentration

matographic analysis are usually extracted using

of 10 mg / ml.

appropriate solvents.

There are many papers on the determination of

2

.2. Instrumentation

carbonyl compounds in aqueous solutions but little
work has been done on their determination in

Gas chromatographic analyses were carried out

alcoholic solutions. Recently, the procedure recom-

with a Perkin-Elmer Auto System XL GC instrument

mended for the determination of aldehydes and

63

equipped with a

Ni ECD system and a split /

ketones, based on their derivatization with PFBHA

splitless injector. The column used was a Rtx-5

and GC with electron capture detection (ECD) has

capillary column (30 m30.32 I.D., 3 mm film

been adapted for alcoholic solutions [13]. Liquid–

thickness). The split / splitless injector and detector

liquid extraction (LLE) with subsequent GC–ECD

temperatures were set at 280 and 250 8C, respective-

permits to determine the C –C

carbonyls at the

1

6

ly. The initial oven temperature was kept at 100 8C

ng / ml level in different spirits and alcoholic bever-

for 1 min, which was increased to 160 8C at 3 8C /

ages.

min, then raised to 220 8C at 5 8C / min and finally to

Solid–phase microextraction (SPME) constitutes a

280 8C at 20 8C / min. The total time run was 36 min.

convenient alternative to commonly used extraction
techniques (especially LLE) because is a simple,
solvent-free, inexpensive, reliable and easily auto-

2

.3. Derivatisation procedure

mated technique [14]. This technique has been
successfully applied for the determination of a wide

PFBHA

reacts

with

the

carbonyl

species

spectrum of analytes in a large variety of matrices

(R COR ) to produce two oxime isomers (Z, E )

1

2

[15–19].

when the alkyl groups R

and R

are different.

1

2

In this work, the possibility of using headspace

Usually, the two isomers can be chromatographically

SPME sampling of oximes formed from corre-

separated [13]. The derivatisation process was car-

sponding carbonyl compounds in reaction with the

ried out in ethanol–water solutions. First, 10 ml of

PFBHA in alcoholic solutions to prior their chro-

an alcoholic solution of carbonyl compounds was

matographic determination has been studied.

placed in a 16-ml vial and then 0.1 ml of PFBHA
solution (10 mg / ml) was added. The vial was
capped with PTFE-faced silicone membrane and

2

. Experimental

heated using a water bath at 45 8C for 1.5 h. After
derivatisation, the sample was cooled to room tem-

2

.1. Reagents and standards

perature and two drops of concentrated sulfuric acid
were added to adjust to pH 2. After shaking, the

The standards of methanal, ethanal, propanal,

sample was ready for SPME experiments. The

propenal (acrolein), butanal, isotutanal (isobutylral-

relatively long time needed for derivatisation in-

dehyde),

pentanal

(valeraldehyde),

2-butenal

fluences the total time for analysis. Recently, a new

(crotonaldehyde), isopentanal (isovaleraldehyde) (all

derivatisation methodology with PFBHA using a

except methanal 97–99%) were purchased from

microwave oven was proposed reducing the reaction

Sigma–Aldrich. Methanal was provided in the form

time by a factor of 50–100 [20].

W

. Wardencki et al. / J. Chromatogr. A 984 (2003) 89–96

91

2

.4. SPME procedure

3

.1. Selection of optimal concentration of alcohol

A SPME holder for manual use and 100 mm

The presence of alcohol (methanol or ethanol) in

PDMS (polydimethylsiloxane) was purchased from

the investigated solutions may act as a co-solvent for

Supelco. This fiber was selected on the basis of

partitioning of carbonyl compounds in the phases

previous experiments with direct SPME of oximes

involved [21,22]. Therefore, the first step was to

from alcoholic solutions [13]. During extraction, the

check the effect of alcohol concentration on ex-

sample was agitated using a magnetic stirrer. Imme-

traction efficiency. The exposure was carried out for

diately after extraction, the fiber was introduced into

the samples with the same amount of carbonyl

the GC injector for 5 min in split mode (1:20). By

compounds (60 mg / l) but with different alcohol

exposing the fiber to the carrier gas stream, the

content (10, 20, 40 and 70%, v / v). Each experiment

analytes were thermally desorbed at 250 8C and

was carried out three times and results were aver-

transferred onto the GC column. Optimal desorption

aged. Fig. 1 presents the peak area of oximes versus

time was found to be 5 min. This time was enough to

ethanol concentration. It may be seen that the

ensure total desorption and no memory effects were

efficiency decreases with an increase in alcohol

observed which was confirmed by desorbing the

content. This suggests that alcohol competes with

same fiber a second time after the initial desorption.

absorption in PDMS film which was suggested
earlier [23–25]. Furthermore, higher concentrations
of alcohol in solution favour the conditions for
forming acetals and hemiacetals. On the other hand,

3

. Results and discussion

it should be noted that high dilution may decrease
the limit of detection for particular aldehydes and

The effects of the main parameters that can affect

acetone. Similar results were obtained when metha-

the SPME process from headspace, i.e. temperature,

nol was used [26]. The direct immersion of the fiber

extraction time, agitation and ionic strength were

in alcoholic solutions also confirmed this conclusion

evaluated. In the case of a solution containing

[26]. As a compromise between efficiency and limit

alcohols, it was also necessary to check the effect of

of detection, an ethanol–water (20:80) solution was

alcohol content for extraction efficiency.

assumed to be optimal.

Fig. 1. Effect of ethanol concentration on extraction efficiency of the oximes at 35 8C from headspace SPME of ethanol–water solutions
(100 mm PDMS fiber, 30 min exposure time, desorption time 5 min at 250 8C).

92

W

. Wardencki et al. / J. Chromatogr. A 984 (2003) 89–96

3

.2. Effect of temperature

constant temperature, all SPME experiments were
conducted at 35 8C.

Extraction temperature controls the diffusion rate

of analytes into the coating. An increase in extraction

3

.3. Effect of extraction time and agitation

temperature causes an increase in the extraction rate
and a simultaneous decrease in the distribution

It is not required to reach the equilibrium by the

constant between the analytes and the fiber. Optimi-

SPME analysis. But exposure time strongly influ-

sation of extraction temperature is generally more

ences the extraction efficiency and eventually affects

important with headspace SPME than when working

the detection limit. Usually the exposure time is a

with direct immersion of the fiber in the liquid

compromise between the analysis time and required

sample. Bao et al. [8] and Cancho et al. [12] used

detection limit. Absorption–time profiles on PDMS

headspace extraction of oximes from aqueous solu-

100-mm fiber in headspace extraction mode were

tions at room temperature. Room temperature is also

generated for each carbonyl compound and are

recommended in the method proposed by the US

presented in Fig. 3. Each data point is the average of

Environmental Protection Agency (EPA) [9].

three independent measurements. For methanal,

The effect of temperature in the extraction was

ethanal, propanal and isobutanal, equilibrium was

investigated varying the temperature between 25 and

reached in 30 min. Higher-chain aldehydes and

45 8C with a constant extraction time. As is shown in

acetone needed longer exposure times. Considering

Fig. 2 an increase in temperature generally improves

this conclusion and duration time for analysis, an

the mobility of carbonyl species through liquid and

extraction time of 30 min was chosen for subsequent

gas phase and better efficiencies were obtained. In

analyses.

the case of acrolein, a decrease in the extraction

Agitation is an important parameter that affects the

yield was observed. This may be caused by instabili-

time profiles. For the analytes that are less volatile,

ty of this unsaturated aldehyde at elevated tempera-

the extraction efficiency is usually notably enhanced

ture. Considering the influence of temperature on the

by stirring because the transfer of the compounds

extraction efficiency (ambient temperature is usually

from liquid solutions to headspace could conceivably

variable in the range of 5 8C) to keep the vials at

be speeded up by agitation. This was confirmed by

Fig. 2. Influence of temperature on detector response area of the oximes formed from carbonyl compounds (concentration level 90 mg / l)
and headspace SPME extraction (30 min) with the 100 mm PDMS fiber from ethanol–water (20:80) solutions.

W

. Wardencki et al. / J. Chromatogr. A 984 (2003) 89–96

93

Fig. 3. Extraction-time profiles (magnetic stirring) for the oximes of carbonyl compounds in 20% ethanol–water solutions by headspace
SPME (100 mm PDMS fiber, 60 mg / l carbonyl compounds concentration, 30 min exposure time, desorption time 5 min at 250 8C).

our experiments, both for direct [26] and headspace

two agitation systems were observed [26]. Therefore,

SPME. Comparing Figs. 3 and 4, it is clear that

a magnetic stirrer was used in the following experi-

extraction efficiently was notably enhanced using a

ments.

magnetic stirrer, especially for higher-chain com-
ponents. In spite of the fact that ultrasonic agitation

3

.4. Salt effect

of the sample should be more effective, because the
two phases are mixed, small differences between the

The addition of salting-out agents improved ex-

Fig. 4. Extraction-time profiles (without stirring) for the oximes of carbonyl compounds in ethanol–water (20:80) solutions by headspace
SPME (100 mm PDMS fiber, 60 mg / l carbonyl compounds concentration, 30-min exposure time, desorption time 5 min at 250 8C).

94

W

. Wardencki et al. / J. Chromatogr. A 984 (2003) 89–96

Fig. 5. Influence of sodium chloride content on detector response area of the oximes (100 mm PDMS fiber, 60 mg / l carbonyl compounds
concentration, 30-min exposure time, desorption time 5 min at 250 8C).

traction efficiency for many analytes in different

3

.5. Calibration

samples. The presence of dissociated ions decreases
the solubility of analytes, which then partition more

Calibration have to be carried out for each com-

readily into the headspace. Furthermore, a salt alters

pound in order to achieve accurate quantitative

the phase boundary enhancing the volatilisation into

results.

the headspace of analytes dissolved in liquid phase.

The linearity of the calibration graphs was tested

The effect of ionic strength on extraction ef-

with at least five calibration points over the expected

ficiency was evaluated by analysing the amount of

concentration range of carbonyl species in alcoholic

carbonyl compounds extracted in sample solutions

beverages. For each concentration level, five in-

containing 0, 10 and 20% (w / v) of sodium chloride

dependent measurements were made. Table 1 pre-

(20% alcohol solutions). These results are shown in

sents the equations for calibration curves and correla-

Fig. 5. As can be seen, the response of oximes

tion coefficients (r) and the relative standard devia-

increased in proportion to the added amount of NaCl.

tions (RSDs) for the compounds tested. The weak

Table 1
Parameters of calibration curves for selected carbonyl compounds in the concentration range from 8 mg / l to 0.32 mg / l using headspace
SPME extraction

Compound

Equation of calibration

Correlation

RSD (%,

curve

coefficient

n53)

Methanal

y514 327 282x24394

0.99825

10.5

Ethanal

y55 743 203x15992

0.99412

13.5

Dimethylketone

y52 403 352x128 727

0.99448

6.6

Propanal

y511 973 054x130 960

0.98917

15.4

Propenal

y59 463 557x120 842

0.97991

5.5

Butanal

y524 311 503x214 208

0.99761

9.8

Isopentanal

y541 180 319x2357 586

0.94337

13.7

2-Butenal

y56 645 899x1141 035

0.89896

11.1

Pentanal

y525 679 309x1206 671

0.96088

14.7

Hexanal

y517 739 371x2126 481

0.99401

11.1

W

. Wardencki et al. / J. Chromatogr. A 984 (2003) 89–96

95

Fig. 6. GC–ECD chromatogram after headspace SPME extraction of oximes after derivatization of carbonyl compounds with PFBHA of
real sample (sample No. 1, vodka, 40%) diluted to 20% ethanol. 15Methanal; 2, 2*5ethanal; 35dimethylketone; 4,4*5propanal;
5,5*5propenal; 6,6*5isobutanal; 7,7*5butanal; 8,8*5isopentanal; 9,9*52-butenal; 10,10*5pentanal; 11,11*5hexanal. Numbers with and
without asterisks: E and Z isomers.

correlation for 2-butenal and isopentanal may be

distilled alcoholic beverages (vodkas) were analysed.

explained by the overlapping of corresponding oxime

Fig. 6 shows a chromatogram of oximes extracted

isomers and using only one peak for quantification.

under optimised conditions described above from

The detection limits were estimated as three times

alcoholic beverages (40%). The initial sample was

the standard deviation of the baseline noise and

diluted to 20% ethanol before the derivatisation and

ranged from 0.05 to 0.5 mg / l of 100% alcohol.

extraction steps. The content of particular com-
pounds was calculated for 1 liter of the 100% spirit
and is given in Table 2. The results obtained showed

3

.6. Real samples

that acetaldehyde was the most abundant aldehyde in
all samples of vodkas, ranging from 0.1 to 0.9 mg / l.

The head space SPME method developed was

It confirms that ethanol oxidation is a dominant

applied to real alcoholic samples. Four samples of

process in beverage production.

Table 2
Content of carbonyl compounds in distilled beverage (vodka, 40% ethanol) determined by the headspace SPME–GC–ECD procedure

Carbonyl

Concentration of carbonyl compounds in distilled alcoholic beverages

compound

(mg / l 100%, n53)

No. 1

No. 2

No. 3

No. 4

Methanal

0.010060.002

0.013060.002

0.010060.002

0.02260.004

Ethanal

0.102060.007

0.183060.013

0.070060.005

0.91360.062

Dimethylketone

0.006060.0006

0.014060.001

0.017060.002

0.11460.004

Propanal

0.001060.0002

0.001060.0002

–

0.08460.008

Propenal

–

–

–

0.11460.004

Butanal

0.005060.0005

0.005060.0005

0.002060.0002

0.008060.0007

Isopentanal

0.015060.002

0.012060.002

0.01060.002

0.03560.004

2-Butenal

0.034060.005

–

–

–

Pentanal

0.006060.0008

0.006060.0008

–

0.03760.005

Hexanal

0.027060.001

0.030060.001

0.019060.001

0.07260.012

96

W

. Wardencki et al. / J. Chromatogr. A 984 (2003) 89–96

[6] A. Yasuhara, T. Shibamoto, J. Chromatogr. 547 (1991) 291.

4

. Conclusions

[7] M.N. Lau, J.D. Ebeler, S.E. Ebeler, Am. J. Enol. Vitic. 50

(1999) 324.

The feasibility of SPME for the analysis of

[8] M. Bao, V. Pantani, O. Griffini, D. Burrini, D. Santiani, K.

carbonyl compounds in different distilled alcoholic

Barbieri, J. Chromatogr. A 809 (1998) 75.

beverages is demonstrated. First, a sample of spirit or

[9] EPA Method 556, Determination of Carbonyl Compounds in

Drinking Water by Pentafluorobenzylohydroxylamine De-

distilled alcoholic beverage is diluted to 20% (v / v)

rivatization and Capillary Gas Chromatography with Elec-

alcohol and carbonyls are derivatised with PFBHA.

tron Capture Detection, EPA, Office of Research and De-

Next, the corresponding oximes are extracted using

velopment National Exposure Research Laboratory, Cincin-

100 mm PDMS fiber by direct immersion into

nati, OH, USA, 1998.

solution at room temperature or from headspace at

[10] J. Nawrocki, I. Kalkowska, A. D©abrowska, J. Chromatogr. A

749 (1996) 157.

35 8C. The analytical characteristic of the headspace

[11] P.A. Martos, J. Pawliszyn, Anal. Chem. 70 (1998) 2311.

SPME method: linearity, precision and limit of

[12] B. Cancho, F. Ventura, M. Galceran, J. Chromatogr. A 943

detection is comparable with direct SPME [26]. But

(2002) 1.

it was proved that the headspace SPME sampling is

´

[13] W. Wardencki, J. Orlita, J. Namiesnik, Fresenius J. Anal.

less sensitive to the matrix effects and prolongs the

Chem. 369 (2001) 661.

[14] J. Pawliszyn, Solid Phase Microextraction: Theory and

life time of the fiber used. Total analysis time is

Practice, Wiley–VCH, New York, 1997.

about 2 h (1.5 h for derivatisation and 30 min for

[15] M. Correia, C. Delerue-Matos, A. Alves, Fresenius J. Anal.

chromatographic analysis).

Chem. 369 (2001) 646.

[16] G. Fitzgerald, K.J. James, K. MacNamara, M.A. Stack, J.

Chromatogr. A 896 (2000) 351.

[17] M. Fernandez, C. Padron, L. Marconi, S. Ghini, R. Colombo,

A

cknowledgements

A.G. Sabatini, S. Girotti, J. Chromatogr. A 922 (2001) 257.

´

[18] T. Gorecki, X. Yu, J.B. Pawliszyn, Analyst 124 (1999) 643.

This research was financially supported by the

[19] N.P. Brunton, D.A. Cronin, F.J. Monahan, R. Durcan, Food

Polish Scientific Committee (Grant 3 TO9A 141

Chem. 68 (2000) 339.

180).

[20] S. Strassnig, T. Wenzl, E.P. Lankmayr, J. Chromatogr. A 891

(2000) 267.

[21] M. Correia, C. Delerue-Matos, A. Alves, J. Chromatogr. A

889 (2000) 59.

R

eferences

[22] R. Eisert, K. Levsen, Fresenius J. Anal. Chem. 351 (1995)

555.

[1] D. Kitts, C. Wu, H. Stich, J. Agric. Food Chem. 41 (1993)

[23] S.E. Ebeler, M.B. Terrien, C. Butzke, J. Sci. Food Agric. 80

2353.

(2000) 625.

[2] V.J. Feron, H.P. Til, F. de Vrijer, R.A. Woutersen, F.R.

[24] D. De la Calle Garcia, S. Maguaghi, M. Reichenbacher, K.

Cassee, P.J. Van Bladeren, Mutat. Res. 259 (1991) 363.

Danzer, J. High Resolut. Chromatogr. 19 (1996) 257.

[3] M. Possanzini, V. DiPalo, Chromatographia 40 (1995) 134.

[25] A.L. Simplicio, L. Vilas Boas, J. Chromatogr. A 833 (1999)

¨

[4] M. Vogel, A. Buldt, U. Karst, Fresenius J. Anal. Chem. 366

35.

(2000) 781.

[26] W. Wardencki, in: Proceedings of the Aldehydes 2001

¨

[5] D.W. Lehmpuhl, J.W. Birks, J. Chromatogr. A 740 (1996) 71.

Conference, Munster, 6–8 June, 2001, p. 16.